Models

Every model you can pick, and the model that answers behind the name. The Runs on column is the provider's own listing for it, as of October 21, 2026. When a model is unavailable, the nearest standard model answers instead and the reply says which one did — the switch is never silent.

Standard models

Included with every plan, unlimited on paid plans. They never use credits.

ModelRuns onPrice
GPT-5 NanoOpenAIGPT-5 Nanoopenai/gpt-5-nanostandard · included
GPT-OSS 20BOpenAIGPT-OSS 20Bopenai/gpt-oss-20bstandard · included
Mistral Small 4MistralMistral Small 4mistralai/mistral-small-4standard · included
Mistral Small 3.2MistralMistral Small 3.2mistralai/mistral-small-3.2standard · included
Llama 3.3 70BMetaMeta Llama 3.3 70Bmeta/llama-3.3-70b-instructstandard · included
Llama 4 ScoutMetaMeta Llama 4 Scoutmeta/llama-4-scoutstandard · included
DeepSeek V4 FlashDeepSeekDeepSeek V4 Flash 0731deepseek/deepseek-v4-flashstandard · included
Gemma 4 26BGoogleGemma 4 26B A4Bgoogle/gemma-4-26b-a4b-itstandard · included
Gemma 4 31BGoogleGemma 4 31Bgoogle/gemma-4-31b-itstandard · included
Nova MicroAmazonNova Microamazon/nova-micro-v1standard · included
Qwen3 Coder 30BAlibabaQwen3 Coder 30Bqwen/qwen3-coder-30b-a3b-instructstandard · included
Qwen3.7 FlashAlibabaQwen3.7 Flashqwen/qwen3.7-flashstandard · included
Nemotron 3.5 LightningNVIDIANVIDIA Nemotron 3.5 Lightningnvidia/nemotron-3.5-lightningstandard · included
GPT-6 LunaOpenAIGPT-6 Lunaopenai/gpt-6-lunastandard · included
GPT-5.6 LunaOpenAIGPT-5.6 Lunaopenai/gpt-5.6-lunastandard · included
GPT-4o MiniOpenAIGPT-4o Miniopenai/gpt-4o-ministandard · included
Gemini 2.5 Flash LiteGoogleGemini 2.5 Flash Litegoogle/gemini-2.5-flash-litestandard · included
Qwen3.8 FlashAlibabaQwen3.8 Flashqwen/qwen3.8-flashstandard · included
Qwen3 VL 30BAlibabaQwen3 VL 30B A3B Instructqwen/qwen3-vl-30b-a3b-instructstandard · included
Qwen3 Coder NextAlibabaQwen3 Coder Nextqwen/qwen3-coder-nextstandard · included
Llama 4 MaverickMetaMeta Llama 4 Maverickmeta/llama-4-maverickstandard · included
DeepSeek V4.1 FlashDeepSeekDeepSeek V4.1 Flashdeepseek/deepseek-v4.1-flashstandard · included
MiMo V2.6 FlashXiaomiMiMo V2.6 Flashxiaomi/mimo-v2.6-flashstandard · included
Step 3.5 FlashStepFunStepFun Step 3.5 Flashstepfun/step-3.5-flashstandard · included
Mercury 2InceptionMercury 2inception/mercury-2standard · included
Hunyuan Hy3TencentHy3tencent/hy3standard · included
Ling 3.1 FlashinclusionAILing 3.1 Flashinclusionai/ling-3.1-flashstandard · included

Included with paid plans

No credits; capped per week.

ModelRuns onPrice
GeminiGoogleGoogle Gemini 2.5 Flash Lite Searchgoogle/gemini-2.5-flash-lite:searchincluded · 300 chats a week

Premium models

A fixed number of credits per message, whatever its length. A stopped or cut-off answer is charged only for what arrived.

ModelRuns onPrice
GPT-5.4 NanoOpenAIGPT-5.4 Nanoopenai/gpt-5.4-nano1 credit / message
GPT-5.4 MiniOpenAIGPT-5.4 Miniopenai/gpt-5.4-mini2 credits / message
GPT-5.4OpenAIGPT-5.4openai/gpt-5.44 credits / message
Claude Haiku 4.5AnthropicClaude Haiku 4.5anthropic/claude-haiku-4.52 credits / message
Claude Sonnet 4.6AnthropicClaude Sonnet 4.6anthropic/claude-sonnet-4.65 credits / message
Claude Sonnet 5AnthropicClaude Sonnet 5anthropic/claude-sonnet-55 credits / message
Gemini 3 FlashGoogleGemini 3 Flash Previewgoogle/gemini-3-flash-preview2 credits / message
Gemini 3.1 ProGoogleGemini 3.1 Pro Previewgoogle/gemini-3.1-pro-preview4 credits / message
Grok 4.20xAIGrok 4.20x-ai/grok-4.205 credits / message
DeepSeek V4 ProDeepSeekDeepSeek V4 Pro 0813deepseek/deepseek-v4-pro3 credits / message
Mistral Large 3MistralMistral Large 3mistralai/mistral-large-31 credit / message
Perplexity SonarPerplexityPerplexity Sonar Fast Searchperplexity/sonar2 credits / message
GPT-5.5 ThinkingOpenAIGPT-5.5openai/gpt-5.510 credits / message
GPT-5.5OpenAIGPT-5.5openai/gpt-5.57 credits / message
Claude Opus 5AnthropicClaude Opus 5anthropic/claude-opus-522 credits / message
Qwen3.7 MaxAlibabaQwen3.7 Maxqwen/qwen3.7-max2 credits / message
Kimi K2.6MoonshotMoonshot Kimi K2.6moonshotai/kimi-k2.63 credits / message
GLM 5.3 FlashZhipuZ.ai GLM-5.3 Flashz-ai/glm-5.3-flash1 credit / message
Claude Haiku 5.5AnthropicClaude Haiku 5.5anthropic/claude-haiku-5.51 credit / message
Claude Sonnet 5.5AnthropicClaude Sonnet 5.5anthropic/claude-sonnet-5.54 credits / message
Claude Opus 5.5AnthropicClaude Opus 5.5anthropic/claude-opus-5.56 credits / message
GPT-6.1 SolOpenAIGPT-6.1 Solopenai/gpt-6.1-sol5 credits / message
GPT-5.6 SolOpenAIGPT-5.6 Solopenai/gpt-5.6-sol3 credits / message
GPT-5.3 CodexOpenAIGPT-5.3 Codexopenai/gpt-5.3-codex5 credits / message
Gemini 3.7 FlashGoogleGemini 3.7 Flashgoogle/gemini-3.7-flash3 credits / message
Gemini 4 ArgonGoogleGemini 4 Argongoogle/gemini-4-argon7 credits / message
Grok 4.6xAIGrok 4.6x-ai/grok-4.64 credits / message
Kimi K3MoonshotKimi K3moonshotai/kimi-k35 credits / message
Mistral Large 4MistralMistral Large 4mistralai/mistral-large-42 credits / message
MiniMax M3MiniMaxMiniMax M3minimax/minimax-m31 credit / message
GLM 5.3ZhipuZ.ai GLM-5.3z-ai/glm-5.34 credits / message
Nemotron 3 UltraNVIDIANVIDIA Nemotron 3 Ultranvidia/nemotron-3-ultra2 credits / message
Command A+CohereCohere Command A+cohere/command-a-plus2 credits / message
Qwen3.8 MaxAlibabaQwen3.8 Maxqwen/qwen3.8-max4 credits / message

How routing works

On paid plans a standard model is asked for by name on the paid endpoint first; a stand-in from a lower-cost tier answers only when it is down, and the reply says so. A provider that is down or out of quota is skipped, and if none can serve the exact model, the nearest one that is up answers. Premium models are only ever served by the paid endpoint under the id shown above, and are refunded automatically when it cannot serve them.